Updated June 24, 2013 at 6:10 pm: In the original version of this article, I mentioned that charts for March 2012 would be added in an update. These charts are now included, and there are minor changes to the text to reflect this.
Recently TTC Chair Karen Stintz and CEO Andy Byford proposed a trial operation of King Street between Shaw and Parliament as a reserved zone for transit vehicles and taxis during the AM peak period. Their concern and claim is that they cannot provide better service without exclusivity over this area.
I believe that this proposal suffers from a lack of detailed knowledge both of the way the line actually operates and the causes of systemic delays (aka “congestion”) to service. The AM peak is much less of a problem than midday, PM peak, evening and weekend operations on the King route. The TTC would do well to concentrate its proposals on areas and problems where there would actually be some benefit. Asking for a total ban on cars rubs motorists (and their political supporters) the wrong way, and needs to be justified by solid data showing how transit would be improved. The TTC has not done this.
Normally when I write a post with analysis of a route’s operations, I prefer to wait until I have more of the data formatted for presentation. In this case, the debate is already underway in the press and in social media. To contribute some technical background, I have started analysis of route 504 King for the months of November 2011, March 2012 and May 2013.
As with past analyses, the information is taken from the TTC’s vehicle monitoring system known as “CIS” which uses GPS to track the location of all vehicles every 20 seconds. The GPS data are converted to a “flattened” view of a route which is a straight line (think of a piece of string pulled out taut). Once this is done, plotting vehicle movements, headways at points and running times between points is greatly simplified. (I plan to publish a separate article in coming weeks for those interest in the details of this process.)
Note that vehicles on 508 Lake Shore are not included in this analysis. Only three inbound trips are scheduled during the AM peak.
Because the proposal addresses a reduction in running time and reduction of congestion in the central area, I will concentrate in this first article on movement of streetcars over the proposed exclusive area. Other sections of the route and aspects of operation (e.g. headway reliability, short turns, time spent serving stops) will be the subject of future articles.
Each of the PDFs linked here contains eight pages. The data show the running times for cars travelling between Sumach (a stop east of Parliament at the Eastern overpass) and Crawford (one stop east of Shaw). This is roughly the location of the proposed reserved transit way. (The points were chosen because they assist in plotting short turns that occur east and west of both locations.)
The first five pages of each set show weekday data for each calendar week of the month. All weekdays are combined on page 6, Saturdays are on page 7, and Sundays are on page 8. (Victoria Day 2013 is included with the Sundays.)
With both westbound files open, it is easy to flip back and forth between companion pages, notably the collected data for all weekdays (page 6). Although this is a busy chart with many data points (one for each vehicle trip), comparing the two shows how the westbound trip from Sumach to Crawford differs in the 18 months span of the three charts.
Although there are considerable differences later in the day, the times up to 9:00 am have not budged very much. Note that the time axis refers to the time a car left Sumach westbound. The trip to Crawford takes 20-30 minutes, and so part or all of any trip after 8:40 will actually occur after 9:00 am when the proposed transit reservation ends.
The times do not peak until after 9:00 am on weekdays corresponding with the point at which parking westbound on King becomes legal and removes substantial capacity from the road. There are individual examples of longer running times on specific days, but these can all be tracked back to delays at a point that may or may not have had anything to do with auto traffic (to the degree that this can be cross-checked to TTC E-Alerts, I will do so in a future update). Systemic congestion shows up when all of the data points swing upward together for all days.
Between fall 2011 and late winter 2012, there was little difference in running times, and if anything the March 2012 data are the better set. However, by May 2013, the whole band of data has moved up the page and it exhibits several interesting features including an early afternoon peak, a much larger PM peak than in 2011, and even a small mid-evening peak corresponding to the time when theatres let out west of Simcoe (this was also visible in 2011).
An important difference between the two years is the amount of construction on King that blocks traffic and the severe congestion at Spadina where all (rather than only some) of the 510 Spadina service is looping via Adelaide, Charlotte and King. Queues for the turn west to north onto Spadina are aggravated by very heavy traffic on Spadina that often blocks the intersection, and by pedestrians crossing east to west where there is no separate priority signal for turning streetcars. (A paid duty policeman has been added in mid June to this location at the TTC’s expense.) This queue can extend back blocks right through the theatre district.
These files include the “all weekday” charts for segments of the route between Sumach and Crawford. Note that the scale on these is double that of the full link from Sumach to Crawford (0-30 minutes versus 0-60) so that the fine details of the shorter segments are more evident. Also, the times shown are for the vehicle time entering the segment, not the times at Crawford which are used in the earlier overall view.
There is little difference in the AM peak values for any of the segments, although the 2013 data do show more of a rise at around 8:00 am near Yonge than the 2011 data. Collectively, these small changes make up the difference in running times seen on the overall chart for the westbound AM peak trips.
A common feature of some of these segments is that the running time does not change much through the day. In other cases, the effect of parking beginning at 9:00 does appear. Where the running times are steady throughout the day, this shows that congestion effects are minimal, and the benefit of completely banning car traffic are dubious.
Conversely, the effect of construction on westbound service from University to Spadina is obvious looking at 2013 versus 2011 and 2012. As the day wears on, running times become quite extended. This is an extremely serious problem on the King route, but the TTC has made no provision for it in the scheduled service, nor has there been much effort to change local traffic regulations on King to compensate. This is a stark contrast to the changes implemented for construction at Bay and Front both in road usage and in the schedule for the Bay bus.
The data for eastbound trips show a comparable story to the westbound data. The data for all three months of weekdays (page 6 of these files) starts out near the “uncongested” value of 20 minutes and rises with a peak corresponding to the onset of parking at 9:00 am. (Times shown are for the start of the eastbound trip at Crawford and so trips before 9:00 will show the effect of congestion further east after 9:00.)
The 2013 data show slightly more midday congestion than 2011, but the PM peaks for both years are roughly the same. The 2011 PM peak is slightly more concentrated in time than 2013.
Looking at the detailed data for each segment, the segment from Crawford to Portland shows more congestion effects over the day in 2013 than in 2011 or 2012. Some of this may be due to construction blocking the curb lane eastbound (as is now the case at Tecumseth, one stop west of Bathurst), but some may also be due simply to increased traffic. This can only be answered with detailed counts, but that would require the existence of “before” data that the city may not have.
Between Portland and Peter, there is slightly more congestion in 2013 than in 2011/12, again possibly due to construction.
Between Peter and University, there is a spike in running times just around 9:00 am, and this, plus congestion through the day, is much more marked in 2013. Similarly, from University to Yonge values are higher in 2013. However, the PM peak for 2011 is tighter and higher than for 2013. Traffic in the core is affected in 2013 by the closing of Front Street for Union Station reconstruction.
From Yonge to Sumach, there is little difference between the two sets of data. Congestion is not an issue in this part of the route.
A Brief Look At Overall Service
In a future article, I will turn to the behaviour of the service overall, but as a preview, here is a chart of the route’s operation on Wednesday, May 15 (the week before Victoria Day weekend).
For those who have not read earlier articles with this type of chart, a brief explanation.
Each line represents one vehicle plotted in location versus time. A full day from 4:00 am until 3:59 the following morning is plotted on eight pages with three hours on each page. Westbound cars are lines sloping up to the right, while eastbound cars slope down.
Locations on the route have been scaled to values between 0 (Broadview Avenue just north of Erindale at Broadview Station) and 1270 (Dundas just north of Edna at Dundas West Station). This corresponds to a one-way trip of 12.7km by comparison the TTC’s Service Summary shows a round trip of 25.62km, or 12.81km one way. The difference is likely due to rounding of the lengths of individual segments of the route used to translate data from GPS to the linear scale used here.
Certain behaviours show up in the charts:
- Between roughly 5:00 and 7:00 am, cars appear briefly westbound from Queen & Broadview. These are running into service via Parliament and Dundas to Broadview Station because the west-to-north curve at Broadview is out of service (the intersection is scheduled for replacement later in 2013).
- Cars appear for a time stationary at Queensway. These are sitting in Roncesvalles Yard logged on for service, but not yet on the street.
- Where a car is stationary in the middle of the route, this is likely a delay. This can be seen at about 8:10 (page 2) where a westbound car (brown line) sits west of Yonge. Note that three other cars (orange, pink, grey) disappear briefly. These have been diverted around the delay via Wellington and vanish from this chart because they are off route. Another delay later on the same page is eastbound south of Bloor around 9:20.
- Dwell times show up as “steps” in the line. The longer the time spent at a stop (or traffic signal) the wider the step.
- Short turns show up clearly as lines that turn back before reaching the terminus. Many of these are visible westbound at Roncesvalles on page 3 (late morning). Some of this activity is in response to accumulated delay during the AM peak.
- Eastbound short turns are not quite as obvious because the cars go off route. There are two variants depending on which way a car travels around the Parliament, Dundas, Broadview loop. At about 10:35, “purple” disappears at Dundas reappearing at King and Parliament at about 10:48. At about 10:55, “dark green” leaves the route at Parliament and reappears westbound at Dundas just after 11:05.
- Congestion shows up as lines whose slope flattens out. Problems westbound to Spadina are clearly evident after 11:00 am and they get worse as the day goes on.
- A few cars have wonky GPS units that cannot handle signal reflections in the core area. 4012, for example, reports a location in various parts of Etobicoke and Peel Region when it is anywhere near an office tower. Such cars vanish and reappear on the charts. Two examples are clearly visible between 22:20 and 22:50 westbound.
- Bunching of cars is evident from closely spaced lines, and this occurs at all locations and times of day. It is quite common for pairs of cars to leave a terminus close to each other, and for short turn cars to re-enter service right beside a through car rather than filling a gap.
- Short turns continue through the late evening. They are not only a peak-hour problem.
These charts show the headways at various points along the route for each direction. As mentioned earlier, the locations were chosen to sample the service between various turnback/diversion points. Each of the 14 pages in the two sets of charts shows the actual headway between vehicles for Wednesday, May 15, 2013. These can be compared with the overall charts of service above.
A trend line is interpolated through the data points to give a general sense of values. In the central part of the route, this tends to lie close to the scheduled headway (i.e. most or all of the trips are operated although their spacing and destinations may not match the schedule). On outer parts of the route where short turns take their effect, the trend lines move away from scheduled values.
This extract gives the headways (among other information) for the King route for the schedule period beginning May 12, 2013.
As in all previous headway analyses, the overwhelming common factor is bunching of cars. A very short headway (a point near the bottom of the graph) is followed by a much wider headway. The scope of these values commonly lies outside of the TTC’s target 3-minute window, although that “window” is meaningless on a route like King where the a car can be running at almost twice the scheduled headway and still be “on time” for statistical reporting purposes.
Also quite striking is the fact that this behaviour occurs right from the terminals rather than accumulating along the route (although a moderately wide swing can grow as the pair of cars travels down the line). Stepping through the pages from east to west, watch the behaviour of the trend line as it moves down (better headways in the central part of the route) and then back up (short turns cause worse headways on outer parts of the line).
There are far more data to plough through in reviewing the King car, but the basic conclusion here is that an AM peak transit reservation from Shaw to Parliament will do little to improve transit operations while creating great annoyance for motorists. There are far worse problems on the King route both with scheduling (inadequate running time and service levels to handle actual conditions on the route), line management (chronic bunching and short turns that may do little to improve service), and traffic management (parking and other factors that steal capacity from the roadway).
The TTC risks using vital political capital and time on a proposal of only modest value while avoiding the much more difficult battles required (both externally and internally) to address all-day problems on the King route.